Easily, a myoelectric arm gets out of control.
But I liked the next one even better. And mine also has such tendencies. They are inherent to the technology. These things really have a mind on their own.
The problems of erroneous activation is extremely well known since many years. There are also studies about it, and anyone that wants to perfect their performance using a myoelectric arm should know about this.
Frequent below-elbow amputee problem: co-activation of electrodes
My problem is that my outer electrode co-activates with elbow flexion. Despite already tricking around.
The problem is known academically as well.
It is termed posture or postural interference, or limb position effect[bibcite key=scheme2010examining,fougner2011resolving,geng2012toward,jiang2013effect,boschmann2013reducing,khushaba2014towards].
The problem lies in the fact that myoelectric electrodes pick up signals from muscle groups that overlap electrically if not also mechanically; so extending or flexing my hand using the strongest stump muscles coincides with an anatomical location where also, electric fields are picked up from adjacent and co-activated muscles that work when flexing or extending the elbow.
So this is not an artifact in any proper sense, but an inherent signal fault.
The myoelectric sensors pick up two signals, ideally: a perfect forearm stump extensor signal and a perfect forearm stump flexor signal.
The best SNR (signal-to-noise-ratio) for controlling prosthetic hand opening and closing usually is obtained somewhere on the forearm where, however, other muscles are located as well; typically, the forearm extensor muscle electrode location coincides with the brachioradialis muscle, and I would suspect the various roles of pronator teres and brachioradialis during elbow extension and flexion in conjunction with their coinnervation of more distal forearm muscles that I find myself unable to fully relax upon elbow extension or flexion, make both of the usually chosen forearm stump electrode location less of a perfectly isolated precision place, but, more of a Bermuda triangle of sorts.
So, electrodes that pick up "hand open" or "close" signals when in fact the elbow is moved do not pick up a "wrong" signal. The signal will undeniably be there.
The problem is that the user of the prosthetic hand at this moment will want to use his elbow, not his hand, and he finds himself unable to provide a naturally occurring easy to accomplish solution. The reason is that we are built to minimal weight for maximal performance, and that means to have (spatially) overlapping muscles that allow to integrate elbow, wrist and hand function.
To avoid this technically, one would have to to identify muscle signals that one can use for prosthetic control that one can avoid activating otherwise. Seeing as if this highly relevant problem is known since almost a decade and absolutely no solution in sight, we have to regard it as one of the inherent and intractable myoelectric problems.
In the following two videos, I do NOT let go of the tickets, or cup, voluntarily.
The hand opens against my will simply by flexing the elbow, and it will remain closed against my will when I extend it (or open, that happened too). Simply, because that is how two electrode myoelectric control really works. It basically screws you.
In the following image sequence (C) Copyright Nicole Kelly) we see how she tries these steps in sequence:
a. grab bottle with extended elbow and arm slightly downward
b. lift arm while slightly extending elbow
This will invariably shift the relative weight of an extensor placed "open" function with activation of both biceps and brachioradialis muscle to a correctly measured "open" signal upon which (engineering wise: correctly!) the hand opens.
The following sequence (image (C) Copyright Nicole Kelly) shows how she:
a. lifts arm at shoulder slightly extended at elbow,
b. tries to close the hand to achieve a grip, but due to over-activation of extensor placed "open" function electrode, can not close the hand.
Again, the myoelectric system correctly identifies an "open" signal, whereas the user also correctly expects the hand to close. The problem lies in the unfulfilled assumption that this is not body powered ; )
More about this here.
Attempts to solution
No one so far escaped the limb position effect, most painfully shown in all gory detail at the 2016's Cybathlon prosthetic arm race event, where all myoelectric users (despite many being professional users, company representatives, key research users with massive training exposure) failed the "hot wire" which only tests limb position effect (and not at all grasp intricacies). Also, other tests were failed but there, presumably, the limb position effect was not the actual cause.
It has not been solved so far. It is 2017, and an answer or solution, provided by research or industry, is not there. So once more the users, us, the arm amputees, need to shine light into this ourselves.
Amputees that have limb positioning effect are blamed with "not mastering their prosthesis" and they should "better train" and "better focus". That is also what Claudia Breidbach of Touch Bionics, who is supposed to offer training to users, told me, "focus, focus more! lift as so!". Before she failed the Cybathlon herself, with the exact same problem of limb positioning effect. So just a bit of focus may not work after all.
Now that is truly irrational: when denying the problem does not work and when the solution is not constructively technical, as engineers may not sufficiently understand it (anatomy can be a real bitch) and researchers do not see any need to solve it (as researchers, themselves, really mostly type up academic papers about studies with "healthy" participants that used some "simulator" while sitting at some desk and never care about their results making it to the end user), then (roughly following the logic of Sherlock Holmes here) placing the burden of this on us, the "irrational amputees", again, remains as the rational answer.
However denying the existence of the limb positioning (or other myoelectric) problems or blaming the amputee does not intrinsically make myoelectric arms more popular. These myoelectric arms thus remain problematic, for obvious reasons. And ultimately, closing their eyes and hysterically riding the dead horses as usual, does not get arm amputees (or industries) any further.
So at first, I assumed this was the way to go:
- First of all, I need to perform a statistic [bibcite key=scheme2010examining].
- Secondly, I do not need "perfect" as long as I can get "good".
- Thirdly, the concept of "manualization" is not one that I can escape. Like, ever.
Early on into trying that and switching between various terminal devices on my myoelectric arm, I found point 1 and 2 were wrong. I found that I need perfect, as there is no way I can accept the hand / gripper to involuntarily opening and dropping stuff, or, locking on to items without being able to let go.
Statistic of limb positioning effect
Really, I found it happens all the freaking time.
Mostly when it is unexpected.
So, carrying expensive equipment is not to be performed with a myoelectric arm - it may open whenever, and then you have it.
You are quite simply better off not wearing a myoelectric arm in order to safely carry items that must not fall down over any distance. In fact, lack of reliable control is one of the reasons why myoelectric arms - despite all that "promise" that goes with them - fall into disgrace far more quickly than one would assume.
That said, you should consider using a bag, or a tray, or a basket, or a box. And you should not make the function of the terminal device that you are wearing any part of a dependent carrying function.
Required degree of perfection
A one out of one hundred drop rate is not acceptable. From where I am standing, the drop rate needs to be practically zero.
Manualization means that the body is turned into a prosthetic arm device control system.
That means that first of all, "compensatory" (better termed "control") motions are necessary to control the prosthetic hand or hook.
More importantly, it means that the functionally best body part is attached to what makes most sense on the prosthetic arm. Seeing as if at least the brachioradial muscle significantly interferes with the forearm stump electrodes every time the elbow is flexed, and seeing as if the electrodes can be swapped, the first step is to swap inside and outside forearm electrode functions. My prosthetic arm is set up so the stump extensors close the prosthetic device, and the stump flexors open it. While that is counterintuitive when reading it, actually working with it is not a problem at all; after a few minutes my brain used this "naturally". The bigger issue is: this is not enough. The brachioradial muscle, when the elbow is flexed, depending on the degree of pronation and supination, also activates the forearm flexor electrode. So one cannot really escape limb positioning effects altogether even though somehow it is a first step.
What really stops limb positioning effect is to stop thinking one can freely move one's arm when wearing a myoelectric prosthesis. One can not. A myoelectric arm is even more demanding on body power, on controlling posture and limb angles, than a body powered arm. With a body powered arm, the control effects are entirely obvious, analog, graded, subtle to forceful, elegant and integrated into a resistance-force-vectoring setup, given that there is cable tension to feel and to push against, or to let go. The myoelectric system is by far more wicked, unintuitive and therefore hard to master as it does requires significant body motion and posture control to avoid erroneous signals such as limb positioning effect.
The true answer to mastering a myoelectric arm therefore is, to first define far more clearly, what body core, shoulder and elbow motion of the arm that works with the myoelectric prosthesis can do for which and for what. And to then incorporate these into what one does.
Myoelectric control error rates
In about four decades, academically published myoelectric arm control error rates remained (a) unacceptably high and (b) became slightly worse [very detailed analysis here]. So the prospects are entirely bleak with regard to ever "truly controlling" this.